Open Interface uses LLM backends to interpret user requests and automate actions on the host computer by simulating keyboard and mouse input. It supports MacOS, Linux, and Windows with Python-based tooling and optional custom LLM configurations.
Collecting history — the radar snapshots this repo daily. The trend line appears after 3 days of data (1 so far).
What it is
Open Interface is a Python-powered tool that self-drives a computer by sending user requests to an LLM backend (such as GPT-4o, Gemini) to determine the steps, then executes those steps via simulated keyboard and mouse input. It can course-correct by providing updated screenshots to the LLM.
How it works
The project describes a loop where the app presents a GUI, captures progress via screenshots, and communicates with an LLM backend to generate instructions. The Core, Interpreter, and Executer components are involved in processing goals, translating them into actions, and performing interactions on the OS.
Getting started
Install/run as a script:
- Install Python 3.12 or newer.
- Clone the repo <code>git clone https://github.com/AmberSahdev/Open-Interface.git</code>
- Enter the directory <code>cd Open-Interface</code>
- Optionally use a Python virtual environment
- <code>pyenv local 3.12.2</code>
- <code>python -m venv .venv</code>
- <code>source .venv/bin/activate</code>
- Install dependencies <code>pip install -r requirements.txt</code>
- Run the app using <code>python app/app.py</code>
Setup sections describe API keys for OpenAI (GPT-4o) and Google Gemini, plus optional custom LLM configuration.
Recent releases
- v0.9.0 (2025-03-16): Added support for free Google Gemini models via open-source contributions; reliability improvements with keyboard hotkey shortcuts; Windows installation notes.
- v0.8.0 (2025-01-12): Big functional and reliability improvements across platforms; fixed MacOS Spotlight launching action; cheaper execution costs.
- v0.7.0 (2024-12-13): UI tweaks and better error handling to prevent crashes; download binaries for OS; setup guidance.
- v0.6.1 (2024-11-09): Bug fix for Custom LLMs like Llava/Llama crashing the app; binaries/downloads guidance.
- v0.6.0 (2024-07-21): Download binaries and setup section guidance; added support.
Traction
Stars: 2707; Forks: 279; Open issues: 25
Behind the repo
Not provided in the README excerpt.
Caveats
License: GPL-3.0. Runtime notes include: cost estimation per LLM request, and requires accessibility permissions on macOS for keyboard/mouse control and screen recording. The setup details mention API keys for OpenAI GPT-4o and Google Gemini. System-specific caveats include potential issues with Open Interface on macOS initial launches and required restarts after API key setup.